PyMOL: Superposition of protein structures

In the next tutorial, we aim to superimpose active (pdb id 7yon) and inactive (7ddz) states of the Y2 receptor. The most significant differences in the structures shall be visualized.

Start using the pml template or work with the GUI and commands.

Introduction and Background

For superimposing two structures, you need to decide on (1) which atoms and which chains or regions to use for the superposition and (2) which method (algorithm). In general, you will usually use Cα atoms for the superposition, as these main chain atoms are best defined and least affected by different main chain and side chain conformations. The region/chain for the superposition depends on the purpose of the superposition.

When analyzing individual differences of local regions (e.g. loops), one would usually superimpose the whole structure. For the comparison of a protein differing in a rigid body motion of domains (or subunits), it is usually better to superimpose one domain (or subunit) to show the difference in orientation of the other domain (or subunit). If one uses the whole protein structure for superposition, no region will give a decent fit and it is difficult to visualize the differences.

PyMOL provides the following commands/methods for superposition.

align Sequence-based alignment of the two structures to determine the corresponding Cα positions and determine the rotation/translation operator to superimpose the structures to achieve a minimum root mean square deviation (rmsd) of the atom positions. The procedure is repeated for several cycles, in which residues exceeding a certain distance threshold (specified in multiples of the rmsd) after superposition are eliminated (e.g. flexible loops). Like this, only the well conserved parts of the structure are used for the superposition.
Use this option for identical structures or structures with medium to high sequence identity (40 % - 100 %, my guess).
super Sequence-independent superposition. Also several cycles are used to exclude regions of local differences.
Use this option for medium to low sequence identity (25 - 40 %).
cealign Uses the cealign algorithm. Use this for very low to insignificant sequence identity (10 - 25 %).

The syntax for the three commands is:

align selection-mobile, selection-reference, cycles=int, cutoff=float
super selection-mobile, selection-reference, cycles=int, cutoff=float
cealign
selection-reference, selection-mobile

cutoff = float: outlier rejection cutoff in RMS {default: 2.0} (RMS: root mean square)
cycles
= int: maximum number of outlier rejection cycles {default: 5}

If you increase the number of cycles or decrease the cutoff, more residues will be excluded from the alignment.

The structural deviations between two superimposed structures are quantified usually by the root mean square deviation (rmsd) in Å. The more atoms one omitted from the superposition in the rejection cycles, the better the rmsd will be. Therefore, an rmsd value should always be specified by the number of atoms used in the calculation and the total number of atoms. Test for the following example the influence of the atom type (all atoms, main chain atoms, Cα atoms) and the number of rejection cycles (start with 0: no rejections) and cutoff.

δi is the distance between atom i and either a reference structure or the mean position of the N equivalent atoms. This is often calculated for the backbone heavy atoms C, N, O, and Cα or sometimes just the Cα atoms (read more about the rmsd value).


Comparison of active and inactive states of the Y2 receptor

Superimpose the active (pdb id 7yon) and inactive (7ddz) states of the Y2 receptor and generate a figure such that the conformational changes are best visible.

Here is one possibility, but find your own first.

Back to PyMOL tutorial main page.